Future groundwater potential mapping using machine learning algorithms and climate change scenarios in Bangladesh

Area of Science:

• Hydrology
• Environmental Science
• Geospatial Analysis

Background:

• Accurate groundwater potential zone mapping is crucial for sustainable water resource management.
• Climate change significantly impacts hydrological systems, necessitating future-oriented assessments.
• Machine learning offers powerful tools for integrating complex environmental data for predictive modeling.

Purpose of the Study:

• To estimate future groundwater potential zones using machine learning algorithms and various climate change scenarios.
• To compare the performance of Artificial Neural Network (ANN), Logistic Model Tree (LMT), and Logistic Regression (LR) for groundwater potential mapping.
• To project groundwater potential zones under different Representative Concentration Pathways (RCPs) for near-future years.

Main Methods:

• Utilized fourteen geospatial and climatic parameters (e.g., slope, rainfall, geology, land use) as inputs for machine learning models.
• Applied and compared ANN, LMT, and LR models to delineate groundwater potential zones, selecting the best model via ROC curve analysis.
• Integrated future precipitation data from RCP scenarios (2.6, 4.5, 6.0, 8.5) to forecast groundwater potential zones for 2025, 2030, 2035, and 2040.

Main Results:

• Artificial Neural Network (ANN) demonstrated superior accuracy (AUC: 0.875) compared to LMT and LR.
• The ANN model identified substantial areas with very high (23.10%) and extremely high (33.50%) groundwater potential.
• Generated sixteen distinct future groundwater potential zone maps across various RCP scenarios and time points.

Conclusions:

• Machine learning, particularly ANN, is highly effective for predicting future groundwater potential zones under climate change.
• The findings provide critical spatial information for national-level water resource planning and evidence-based decision-making.
• Proactive water management strategies are essential to address the projected changes in groundwater availability.